Database New Added 8 September 2026

Redis Cluster: CROSSSLOT Keys in request don't hash to the same slot

A multi key command touched keys living on different shards. Redis Cluster splits the keyspace into 16384 slots by hashing the key, and commands like MGET, transactions and Lua scripts must stay inside one slot, so code that worked on a single node breaks on a cluster.

Quick fix

Read the commands before running them. Anything that restarts a service, deletes data or changes permissions should be tried on a non-production system first.

Quick fix

# Force related keys into one slot with a hash tag
SET {user:42}:profile ...
SET {user:42}:sessions ...
MGET {user:42}:profile {user:42}:sessions     # same slot, allowed

# Check where a key lives
redis-cli -c cluster keyslot 'user:42:profile'

# Or issue one command per key and pipeline them
pipe = client.pipeline(transaction=False)
for k in keys: pipe.get(k)
pipe.execute()

How to diagnose Database errors

Database errors group into connection exhaustion, lock contention and deadlocks, constraint violations, and resource limits. Connection errors under load are the most misdiagnosed: "too many connections" is almost never fixed by raising max_connections, because each connection costs memory. It is fixed by putting a pooler in front and finding the code path that leaks connections.

If the quick fix above does not resolve it, work through these steps. They apply to this whole class of error, not just to this one message, which is usually what saves the time.

Look at what the server currently sees: SELECT * FROM pg_stat_activity in PostgreSQL, SHOW FULL PROCESSLIST in MySQL. Sessions sitting in idle in transaction are the usual culprit behind connection exhaustion.
For deadlocks, read the deadlock report the database logs. It names both transactions and the exact lock order. The fix is nearly always to make all code paths acquire locks in the same order.
Check whether the error is from the database or from a pooler. PgBouncer, ProxySQL and RDS Proxy return their own errors that look database-native but need pooler-side fixes.
Run EXPLAIN (ANALYZE, BUFFERS) on slow queries. Lock-wait timeouts are often just very slow queries holding locks longer than they should.
Verify you are connected to the writer, not a read replica. READONLY and "cannot execute in a read-only transaction" errors mean traffic is reaching the wrong endpoint.

Tools worth reaching for

pg_stat_activity
SHOW ENGINE INNODB STATUS
EXPLAIN ANALYZE
slow query log
pgbadger

Authoritative references

Primary documentation for this error, worth reading before applying any fix in production.

redis.io

Related Database errors

See all 41 Database errors →

Browse other categories

Something missing or wrong?

This entry is maintained by hand. If the fix is out of date, incomplete, or you have a better one, email a correction and it will be reviewed.